Along with the enhancement of importance of data used in computer systems, storage operation has been developed where data is multiplexed and retained in multiple storage subsystems located at multiple sites, in order to prepare for disasters such as earth quakes and fire. Methods for multiplexing data among multiple storage subsystems include, for example, a multiplex output method where a host computer writes identical data into multiple volumes, or a remote copy method for multiplexing data among multiple storage subsystems.
According to the remote copy method, different host computers connect to storage subsystems located at different sites, respectively. Usually, the host computer located at a primary site accesses the storage subsystem located at the primary site to carry out operations, and the storage subsystem at the primary site transfers update data to the storage subsystem located at a secondary site, wherein the host computer at the secondary site will not perform output to the storage located at the secondary site. This form of operation is called an active-standby configuration. When disaster occurs at the primary site, the operation of the host computer is handed over to the secondary site, so that operation can be continued by accessing the storage subsystem located at the secondary site.
Regarding the configurations of connection between host computers and storage subsystems, one possible configuration enables multiple host computers to share the same volume in a single storage subsystem, and another possible configuration enables a single host computer to access multiple storage subsystems.
There are two types of systems in the configuration where multiple host computers access the same volume within a single storage system, one of which is a system where a clustering technique is adopted in the host computers, and the other one is a system utilizing an access exclusion technique within the storage subsystems. In the system where a clustering technique is adopted by host computers, when one host computer fails, one of the other remaining host computers takes over the operation. At this time, competition of volume accesses is generally subjected to exclusive control among host computers.
On the other hand, there are two types of methods in the system where access exclusion technique within storage subsystems is adopted, one of which is a method of reserving a usable host computer in volume units, and a method of simultaneously using different access ranges of a single volume from multiple host computers.
For example, patent literature 1 discloses an art in which information concerning extent (extent range) of an input/output processing request issued by a host processor is stored to thereby enable a storage control device to perform exclusive control on an extent-by-extent basis. If the host processor is a mainframe, the extent range is designated in the parameter of a beginning command of a command chain in which channel commands are transmitted sequentially. According to this system, exclusive sharing of volumes among storage control systems can be realized based on the extent range of a beginning command, without having to perform exclusive control of access ranges of volumes among multiple host processors.
Recently, along with the increase of data capacity, the number of storage subsystems located within a single site is increasing. Management operation becomes complex in a large-scale storage system, so that patent literature 2 discloses a storage virtualization technique in which identical storage subsystem identifiers are provided to multiple storage subsystems. Based on this storage virtualization technique, the number of storage subsystem identifications being the target of management can be reduced.
Patent literature 3 discloses an active-active configuration in which identical identifiers are provided to volumes within multiple storage subsystems, wherein a host computer is enabled to achieve the same access results by accessing any one of the multiple storage subsystems. According to patent literature 3, there is only one host computer, and the host computer issues the subsequent input/output request after the previous input/output processing is completed.